Matthew Biju

Matthew Biju

Contact details — on requestProof of Work — on request

Experience

S

AI Engineer

SAP · Apr 2025 – Present

Not yet confirmed
  • • Built and deployed production-grade multi-agent LLM systems using LangGraph/ReAct with 12+ tools, incorporating end-to-end observability through LangSmith, structured output validation, and prompt injection/PII guardrails to improve reliability and reduce manual research effort.
  • • Developed KPI analytics agent systems using LangGraph Deep Agents, featuring sub-agent orchestration, DuckDB-backed MCP servers, and human-in-the-loop approval workflows for business insight generation.
  • • Designed and implemented RAG and document intelligence pipelines using pgvector and HNSW indexing, reranking strategies, and prompt-chained slide generation workflows.
  • • Built multimodal GenAI pipelines exposed through MCP servers, powered by Python and Go microservices communicating over gRPC, and integrated them into customer-facing product experiences.
  • • Delivered full-stack product features using React, TypeScript, and Next.js, including rich-text chart embedding, real-time image composition with Fabric.js, and AI-powered content creation workflows.
U

AI Engineer

Undisclosed Company · Jan 2024 – Present

Not yet confirmed
  • Delivered a production multi-agent LLM system in LangGraph using the ReAct pattern with an orchestrator/worker architecture to fully automate enterprise business research, coordinating 12+ tools through stateful agent memory, dynamic tool routing, and iterative self-correction loops with context-aware reasoning, achieving an 83% reduction in manual research effort.
  • Instrumented end-to-end agent observability, tracing, and prompt evaluation across the multi-agent pipeline via LangSmith, enabling real-time debugging of chain execution, latency profiling, and systematic prompt quality assessment, cutting mean-time-to-debug agent failures by ~60%.
  • Architected a Retrieval-Augmented Generation (RAG) system for in-app semantic search over user-uploaded documents, applying recursive document chunking with OpenAI text-embedding models for dense vector representation, pgvector as the vector store, HNSW indexing for sub-100 ms approximate nearest-neighbour lookup, and a cross-encoder reranking stage with retrieval-relevance evaluation, reducing average query latency.
  • Built an end-to-end document intelligence and presentation generation pipeline that transforms user prompts into structured slide decks via Vector-less RAG (PageIndex) covering document indexing, context-aware prompt chaining for outline and slide generation, and JSON-schema rendering through Reveal.js for interactive browser-based output, generating 10-15 slide decks cutting manual slide-prep effort by ~70%.
  • Engineered an analytics and KPI dashboard system using LangGraph Deep Agents, where a planner orchestrates spawned subagents for SQL generation, chart rendering, and insight summarisation with isolated context windows, leveraging the virtual file system for intermediate artifact persistence and custom middleware for context summarisation and human-in-the-loop approval; ingests user-uploaded CSVs through MCP servers backed by DuckDB and turns raw data into KPI metrics, charts, and actionable business insights with sub-second SQL query response on CSVs of 10K+ rows.
  • Drove frontend development across the product suite using React, TypeScript, and Next.js, delivering 20+ reusable components adopted across 3 product surfaces, achieving WCAG 2.1 AA compliance and responsive layouts, and collaborating with UX designers to translate Figma specifications into production-ready interfaces.
S

Associate AI Engineer

SAP · Jul 2023 – Mar 2025

Not yet confirmed
U

AI Engineer

Undisclosed Company · Jul 2023 – Jan 2024

Not yet confirmed
  • Exposed GenAI agent capabilities as a Model Context Protocol (MCP) server, providing standardised tool-calling endpoints for prompt-based image and video generation; enabled consistent structured LLM outputs and simplified integration for downstream agentic workflows and third-party consumers.
  • Engineered a multimodal content generation pipeline integrating AWS Titan for image generation and AWS Nova for video synthesis, orchestrating Python and Go microservices via gRPC with Protocol Buffers for type-safe inter-service communication and Redis for distributed state management and caching.

Skills 0 proven through work

Also works with

Technical Solution DesignJSON-Based Dynamic UI RenderingCSSHTMLStakeholder ManagementSoftware Development PracticesProject Architecture PlanningLegacy Code AnalysisRequirements AnalysisIncremental Code MigrationPrompt EngineeringJSON Data HandlingLLM Output Verification and ValidationObject Storage ManagementAWS DeploymentLinux Shell ScriptingFile System ManagementThird-Party Library IntegrationWorkflow State ManagementGraph-Based Dependency ModelingFront-end DevelopmentJavaScript DevelopmentReact DevelopmentFastAPI DevelopmentPython DevelopmentLangGraph Framework UsageGenerative AI ApplicationAnalytical Report CreationAPI IntegrationAI Agent DevelopmentAI Workflow AutomationApplication DeploymentMulti-Agent System IntegrationLLM Application Integration

Proof of Work

Proof of Work

Matthew shares this with people who ask. You'll hear back either way.

Education

Bachelor of Technology, Electronics and Communication Engineering

National Institute of Technology Calicut (NIT Calicut) · 2019 — 2023

Contact details

Contact details

Matthew shares this with people who ask. You'll hear back either way.

Ask about this candidate

AI responses may contain errors. Proof status and source information are provided by Proof of Skill.

A Proof CV — one profile, kept current, with the proof attached. Back to the portal